Igor pointed out that the analyzer reports that the F1s are sometimes missing the reference time. A look at the raw data confirms this. To try to diagnose this, the trigger to the F1s was delayed by 115ns and a second copy of the reference time was sent to the F1 in slot 17 (same channel 33 as in slot 15). This will help to reveal if a) the reference time was too close to the trigger, b) the F1 is inefficient, or c) the reference time is simple missing for some events. This change is effective as of run 1144. These changes were made just before run 1144.